What problem does it solve? Turning accepted architecture into work that independent coding agents can execute is hard: plans are often vague, ignore existing design decisions, skip external research, or conflict with parallel workstreams. This Skill produces precise, executable implementation plans grounded in repository evidence, accepted Design documents, and verified external sources, then registers them in a dependency-aware execution graph. ## Core Features & Use Cases - Scope-first planning: An explicit user scope defines the planning boundary, keeping repository inspection and research proportional to the requested work. - Mandatory external research: Every substantive plan requires opened, authoritative web sources recorded in a research register tied to concrete decisions. - Execution graph integration: Plans are registered via MINE MCP tools or the JSON CLI, released from DRAFT to READY or BLOCKED, and validated against the graph. - Parallel execution design: Work packages get disjoint write scopes, serial owners for shared files, and a parallel execution protocol when lanes run concurrently. - Use Case: After mine-arch accepts a design for session storage, invoke this Skill with that scope to produce a numbered plan under docs/plan/ with work packages, verification matrices, and a registered graph node ready for mine-plan-exec. ## Quick Start Ask the agent to create an implementation plan for a specific scoped feature or change, for example: create a MINE plan for adding session storage and recovery based on the accepted design.
